Click on any phrase to play the video from that point.
[Adobe TV Presents]
[Jason Levine]
[Karl Soule]
[Short and Suite] [On the Road with Jason and Karl]
Hello everyone, and welcome to Short and Suite.
My name is Jason Levine, Worldwide Principal Evangelist
for Adobe's digital video and audio tools, and in this episode,
we're going to take a look at AGC repair from DSLR cameras.
So, what exactly is that?
Well, AGC, of course, is automatic gain control,
and this is something that's implemented on all of your DSLRs by default
to prevent signals from distorting or digitally clipping,
and this is okay, and this is a good thing, because it prevents your audio
from being completely unusable.
The problem is is that it tends to also introduce artifacts that you don't want.
Now, the good news is, in many cases, it can be repaired.
And it can be repaired quite seamlessly quite easily in Adobe Audition CS5.5
with just a couple cool little steps and highlighting a couple of cool features
that we have in there which you've probably never seen,
or perhaps you've stumbled upon but weren't sure what to do with them.
So, let's take a look, and let's show you how we do that.
So, what I have here is some footage that I shot on a recent trip to Egypt
with my colleagues Greg Rewis and Terry White.
And we have an Egyptian qanun player, which is like a dulcimer basically,
all shot, again, with my Canon 7D, leveraging the onboard mic.
So typically, I'll use an external microphone, something like a Rode
or again, even a standard dynamic or condenser mic plugged into the camera.
This is actually the onboard microphone, which is not great,
definitely prone to distortions, and again, it's got that AGC acting on it all the time.
Now, some cameras allow you to disable it.
Unfortunately, at the moment, the 7D doesn't, so we're stuck with what we have.
So, if you take a look and listen to this for a couple of minutes,
you'll kind of hear what's going on.
Take a listen.
[♪ Music ♪]
Now, you might not hear any great amount of distortion,
but if you were to take a look at the level meters,
let's play a couple seconds of that again.
[♪ Music ♪]
The level meters are consistently peaking in the red, right?
And that's not good. We don't want that.
Now, of course, you could, within Premiere, just bring or normalize down
or just bring the overall volume of everything down.
The problem is everything's already limited. It's already flatlined.
There's no dynamics to this.
It's just compressed, kind of in your face, doesn't sound particularly good,
and it's going to sound even worse as you continue to export this,
whether you're going down--if you're going to audio via MP3
or going to any kind of tablet device or online.
You really should fix these types of distortions before you do anything else.
And now there's a very simple way to do that directly from the Premiere Pro timeline.
So, from within Premiere, if I want to send this entire sequence--
and by the way, the sequence already contains a couple additional tracks of audio
as well as some video--we can send all that to Audition too.
From under the Edit menu, we can go to "Edit in Adobe Audition, Sequence."
Now, in this dialog, you have a couple of different options.
First, let's give this a name, and I'll call this "QanunWARP-ATV."
We're going to set the entire sequence here,
and then we'll add handles, and you can see we can export the preview video if we want.
We're going to go ahead and do that.
We also have the ability to render any audio clip effects that we've already added in Premiere,
so these will actually be rendered to the audio files, but we don't have any.
More importantly, "Send Clip Volume Keyframe Metadata."
So, you may notice just below here that I actually have some
non-destructive fades on my audio files.
These will be brought over to Audition, and it'll retain their non-destructive nature
so that you can readjust these curves before sending them back to Premiere.
That's very cool. And then lastly, the option to actually open this in Audition.
Of course, once it exports, now I want to see everything.
So, while I'm talking you can see this is happening.
And by the way, once again, the speed of Mercury,
the speed of Premiere Pro CS5.5 to export this--we're done.
Thirty seconds with multiple tracks exported, sent over to Audition, loaded.
Let's pull up our video panel here so you can actually see what's happening.
There it is. Shrink this one up a bit and hit play.
[♪ Music ♪]
We're there. Simple, right?
And just like I said, notice these fade handles here.
Notice non-destructive fades.
Those are the same fades that we just brought over from Premiere Pro
preserved as non-destructive fades in Audition.
That's really awesome, right?
So again, now, what we want to focus on is this cam audio mono track.
It's actually dual-channel mono.
It's stereo, but it's the same audio on the left and right coming off the camera microphone.
So, if I double-click on this to bring it into the Edit view or the Waveform view
as it's now called, right away, even if you don't know a lot about audio,
you can just see that these peaks are hitting, effectively, the ceiling.
Now, this is where we're going to use a couple of panels,
two of which are my absolute favorite in Audition,
the first one being the Amplitude Statistics panel.
Let's go ahead and open this up. Again, this is a dockable panel.
You can place this wherever you'd like. Let's go ahead and dock it over here.
What's great about this is this panel is going to tell you everything you need to know
about your audio, including where it actually peaks,
the actual peak level, the RMS level, the loudness levels,
and then if you have any clipped samples or any DC offset introduced,
which you might not hear right away, but it's still there.
Let's go ahead and click "Scan."
Again, mere seconds to scan the audio,
and as promised via AGC, peak amplitude is 0.
Now, it doesn't show that it's +1.23, +5.
It wouldn't. That would be completely distorted.
So, the AGC is working. It's doing its job.
But as mentioned, that doesn't mean that you're still not introducing
possibly clipped samples.
And sure enough, phrased just like that, "Possibly Clipped Samples," 632.
Now, this is 48 K, 48,000 samples per second.
Do 632 clipped samples in a 30-second clip actually matter?
Yes. Yes, they do.
You don't want to leave those there because if nothing else,
you're also preventing yourself or you're preventing this clip
from actually showcasing any kind of dynamic material, dynamic media--
and by dynamics, I mean audio dynamics--that might exist there.
We've got percussion. We've got all these things.
There are actually changes in volume, and you're just not hearing them.
Now, someone often said to me "Well, again, can't I just drop the level of everything?"
And yes, you can, and if you look at our heads-up display here
where we have this control over volume, so this is just globally making volume changes,
again, everything is already peaking at 0.
If I go ahead and drop this down to around -3 dB--
let's go ahead and move that--so yes, now we are not going to have any clipped--
we're not going to have anything hitting the red there.
But everything is still flatlined.
This is the crew-cut syndrome.
That's not good. It's still limited.
It's still compressed. It still is going to sound not so great.
So, we don't want to do that. That's not helping anything.
All that's doing is saying it's not actually going to clip in the red on playback.
But I actually want to fix what's wrong here.
This is where this gets super cool, and this is why people often have
referred to Audition as the After Effects of audio, because of these little things
that you can do in all these different analysis panels.
So, let's come over here.
Let's make this one a bit bigger, and we're going to bounce over to the Diagnostics panel,
and in the Diagnostics panel, we house 2 different effects:
the DeClicker and the DeClipper.
And because we have clipped samples, we're going to implement the DeClipper.
Go ahead and choose that and click "Scan."
Scanned the whole thing.
Now again, you can scroll through these here.
The window is very small at the moment.
Doesn't matter. What you can see is that we have 64 problems detected.
These are the problems that are causing all of these weird things to happen.
They're causing those clipped samples.
Do you need to know anything about how to repair clipped samples?
No. All you have to do is click "Repair All," click it, and it looks exactly the same.
Weird. Ah, you know why?
Because we're working in 32-bit floating point.
Now, in audio when you talk about bit depth,
you have dynamic ranges related to the bit depth.
At 16-bit, which is CD, CD quality,
you have roughly a 96-decibel range at 16-bit.
At 24-bit, which is DVD standard, you have approximately 140-decibel range.
And in 32-bit floating point, which is the standard operating method inside of Audition,
you have a theoretical dynamic range of 1,500 decibels.
So, we just repaired a bunch of clipped transients,
but we can't see them because they're currently out of range.
So, here's the tip.
Effects, Amplitude, Normalize.
And you are going to do what's known as a downward normalize
to -.5, half of a dB below 0.
In theory, you could go down to -.1.
The truth is, your ear doesn't perceive a change in volume that's any less
than a dB and a half anyway, but -.5 is comfortable enough.
Let's go ahead and click "Okay" and watch and behold.
Whoa. It's not flat-topped.
It actually appears to have dynamic peaks.
And if you take a look here, can you actually see these transient peaks are--
this is actually performing a crescendo.
It's getting louder. That's the percussion.
That's the tambourine getting louder. Take a listen.
[♪ Music ♪]
It's actually crescendoing.
We just restored an audio crescendo on something that was flat-topped moments ago.
If that doesn't excite you, I don't know what will.
It's really exciting, right?
Now, is this going to solve every problem?
No, of course not. It can't, right?
But the idea is, it can solve most problems
and very quickly and truly fast in a fast fashion
repair those artifacts, repair the clipping from AGC
that can destroy an otherwise great take.
So now, when I go back to my amplitude statistics,
and you'll see that it's indicating here--this is telling us that changes have been made.
Let's rescan that and peak amplitude now -.55,
possibly clipped samples 0.
And now we have wonderfully dynamic content
that isn't going to clip, that actually is dynamic, that changes in volume.
It's going to sound better, and now we could perform other operations,
either destructively in the Waveform view or non-destructively in the Mixer view,
and then send all that back to the multitrack, and that's exactly what we're going to do
and send it to Premiere.
So again, from within here, we can choose whatever is it that we want to do.
We could add any number of effects.
We're not going to do that right now.
I just want to get this back to Premiere Pro to once again showcase
just how unbelievably fast this application is,
especially when you work together with Premiere.
So, from the Multitrack menu, we can choose "Export to Adobe Premiere Pro."
Now, something else to point out.
If you're using Audition with, say, Final Cut or another NLE or another DAW,
another audio workstation application, you'll notice under the File Export menu
that you have options here to go out to FCP XML interchange or OMF.
OMF is the standard. People have been asking for this for years in Audition.
Now you have that option.
Import and export OMF, import and export Final Cut XML.
In our case, we're going to go directly back to Premiere Pro.
Let's go ahead and choose that, where now you can see it automatically names it for us,
sample rate, and now you have the option to send things back
as individual stems or as a stereo mix.
And now, this is entirely up to you.
Typically, if you're going to send all of your audio with a video reference from Premiere,
you'll probably return a stereo mix because you can always go back.
You can always edit original and modify all the original pieces,
all the original audio files.
However, if you choose Mixdown, you can choose Mono, Stereo or 5.1.
For me, just to showcase the speed of all of this,
I'm going to choose to export each track as a stem and open it in Premiere.
Are you ready? Again, we've got 30 seconds here, multiple tracks.
Click "Export." One, two, it's done.
Choose where you want it to go and blip.
We just exported 30 seconds of multiple tracks in 3 seconds.
It's that simple.
Now, of course, it brings it back with everything enabled,
including the old audio, so if we go ahead and turn off the old audio clips,
you can see we've got the new ones here.
Go ahead and click "Play."
[♪ Music ♪]
And you can see already that obviously these are the new files
because we don't see any clipping, any red, anything,
and it already sounds better right inside the meters in Premiere.
[♪ Music ♪]
You'll also notice that now just by repairing the AGC artifacts
we can hear the voice-over that much better, and we haven't even remixed the voice-over.
It's that simple. It's so cool.
And again, you can take any clip from any camera,
send it to Audition and begin repairing it, begin tweaking it
and just make it a little bit better with a few simple techniques.
So, for Short and Suite, my name is Jason Levine.
We'll see you next time. Thank you.
[Adobe TV Productions] [TV.Adobe.com]
